The design and implementation of computerised visual inspection machines are labour intensive. Most of the visual inspection systems have been produced in solitude without any well-organised strategy which has led to the purpose of designing rigid customized solutions which require very huge construction costs.
Please refer the link https://www.sipotek.net/ to know more about the designs of the visual inspection system.
The design of Automated Visual Inspection system is divided into two-stage:
- The modelling or training phase
- Inspection Phase
- Image acquisition: It consists of image pre-processing, sensing, and providing the spatial representation of an object and digitalization.
- Feature detection: It is the method by which a primary image is converted into characteristics, which contains a large amount of data to be used for inspection.
- Modeling: Chosen features together with predefined stipulations are used to produce a specific object design, described in a standard data format.
- Analysis: In the inspection phase, the deserted characteristics are chosen by a filtering method to decrease the feature set. A feature is identified if a specific decision rule is capable of assigning it as referring to an assigned subset of target features.
- Matching: Inspection profits by meeting the data structure induced from the perceived picture to the model data structure. Usually, this measure is not insulated from the analysis.
The selection of illumination designs, digitizers, optics and sensors is extremely reliant on the spectrum of transmission of an object, the smallest object characteristic size, the shape of object and speed of movement. Due to a large number of various factors and imaging difficulty, the model acquisition plan is an innovative effort rather than a quantitative model. These sensors can be in the form of CCD cameras, ultrasound or X-Ray camera. A well-designed sensing method reduces noise and stops object movement.
Different strategies for designing visual inspection systems:
- Bottom-up Design: These visual inspection system manufacturersgive a smart solution for a particular problem by unlocking it case-by-case in an improvised manner. This works on a particular obstacle with the demonstrative method. In the beginning, there is an interactive scientific procedure stage, during which given datasets have experimented with real tools. In this stage, all factors are assessed and improved in order to resolve the assigned job in the most suitable way and then the application code is modified based on analysis and experiments. The benefit of this procedure is that the solution is compressed, simple and enough to solve the particular problem. The basic problem of this active solution lies in determining a very specific case. Each application, potential methods and enduring tools have to be examined to find the most suitable solution to solve the problem.
The review process gained in the specific case can rarely be reused. In case of the layout of the object is to be inspected is slightly changed a complete re-design of the inspection system will be necessary.
- Top-down design: This design starts with the description of the problem space in which the particular difficulty is buried. The problem space can be defined by an ideal expression which incorporates both the likely inputs and outputs. The investigation successively improves the concept until operators can be implemented on the information. Such methods can get very complicated since they normally have several data structures with a number of factors that require solving each specific issue.